Kinetic Effects of Increased Proton Transfer Distance on Proton

Publication Date (Web): September 15, 2011 .... Proton-Coupled Electron Transfer Reactions with Photometric Bases Reveal Free Energy Relationships for...
1 downloads 0 Views 2MB Size
ARTICLE pubs.acs.org/JACS

Kinetic Effects of Increased Proton Transfer Distance on Proton-Coupled Oxidations of Phenol-Amines Todd F. Markle,*,§ Ian J. Rhile,† and James M. Mayer* Department of Chemistry, Box 351700, University of Washington, Seattle, Washington 98195-1700, United States

bS Supporting Information ABSTRACT: To test the effect of varying the proton donoracceptor distance in proton-coupled electron transfer (PCET) reactions, the oxidation of a bicyclic amino-indanol (2) is compared with that of a closely related phenol with an ortho CPh2NH2 substituent (1). Spectroscopic, structural, thermochemical, and computational studies show that the two amino-phenols are very similar, except that the O 3 3 3 N distance (dON) is >0.1 Å longer in 2 than in 1. The difference in dON is 0.13 ( 0.03 Å from X-ray crystallography and 0.165 Å from DFT calculations. Oxidations of these phenols by outer-sphere oxidants yield distonic radical cations •OArNH3+ by concerted protonelectron transfer (CPET). Simple tunneling and classical kinetic models both predict that the longer donoracceptor distance in 2 should lead to slower reactions, by ca. 2 orders of magnitude, as well as larger H/D kinetic isotope effects (KIEs). However, kinetic studies show that the compound with the longer proton-transfer distance, 2, exhibits smaller KIEs and has rate constants that are quite close to those of 1. For example, the oxidation of 2 by the triarylamminium radical cation N(C6H4OMe)3•+ (3a+) occurs at (1.4 ( 0.1)  104 M1 s1, only a factor of 2 slower than the closely related reaction of 1 with N(C6H4OMe)2(C6H4Br)•+ (3b+). This difference in rate constants is well accounted for by the slightly different free energies of reaction: ΔG (2 + 3a+) = +0.078 V versus ΔG (1 + 3b+) = +0.04 V. The two phenol-amines do display some subtle kinetic differences: for instance, compound 2 has a shallower dependence of CPET rate constants on driving force (Brønsted α, Δ ln(k)/Δ ln(Keq)). These results show that the simple tunneling model is not a good predictor of the effect of proton donoracceptor distance on concerted-electron transfer reactions involving strongly hydrogen-bonded systems. Computational analysis of the observed similarity of the two phenols emphasizes the importance of the highly anharmonic O 3 3 3 H 3 3 3 N potential energy surface and the influence of proton vibrational excited states.

’ INTRODUCTION Reactions that involve transfer of proton(s) and electron(s) are important in a wide variety of chemical and biological systems, from enzymatic reactions to electrocatalysts for important energy processes such as water oxidation to dioxygen.14 The most studied class of such proton-coupled electron transfer (PCET) reactions involve the transfer of one electron and one proton in the same kinetic step, called concerted protonelectron transfer (CPET).5 Understanding the parameters which affect CPET and other hydrogen transfer reactivity will be broadly valuable. Studies of CPET have been aided by parallels with the Marcus theory of outer-sphere electron transfer (ET), which relates reaction rates with driving force.6 Marcus-type treatments have been used by experimentalists to explain CPET results,1,7 and most of the more extensive theoretical treatments have used a Marcus-type approach as their starting point.8 The greater mass of the proton relative to the electron requires that hydrogen transfers occur over much smaller distances than ET, making the proton transfer (PT) distance a key parameter in CPET. Nuclear tunneling is often important in these processes and is highly dependent on the distance the proton must transfer. One simple tunneling model predicts that an increase of 0.1 Å should lead to a 30-fold decrease in rate.9 Kinetic isotope effects r 2011 American Chemical Society

(KIEs) of deuterium (and tritium) substitution are valuable tools in the study of these processes. To a first approximation, the more massive nuclei have a steeper distance dependence of tunneling probabilities. The PT distance has received particular attention in the study of biological hydrogen atom transfer (HAT), which is one type of CPET process. For example, HAT in lipoxygenase enzymes, from an allylic CH bond to the iron(III)-hydroxo active site, has been of interest since the discovery of hydrogen/deuterium kinetic isotope effects (KIEs) as large as 81.10 This large value indicates that nuclear tunneling is important.10,11 The temperature dependence of the KIE and other data indicate that vibrations which modulate the PT distance are crucial to the action of the enzyme.12 In a series of detailed computational studies, Hammes-Schiffer and co-workers have found that HAT occurs primarily from configurations in which the proton donoracceptor distance has been compressed to 2.69 Å, substantially shorter than the equilibrium distance of 2.88 Å.13 Optimization of these motions in the protein and the resulting effects on proton tunneling have been suggested to be key factors in enzyme catalysis.2,3,13,14 Received: June 19, 2011 Published: September 15, 2011 17341

dx.doi.org/10.1021/ja2056853 | J. Am. Chem. Soc. 2011, 133, 17341–17352

Journal of the American Chemical Society

ARTICLE

Scheme 1. Phenol-amines 1 and 2 and Their Concerted Proton-Electron Transfer Oxidations by Triarylamminium Cations (3+)

Small molecule phenol-base compounds have proven to be valuable model systems for the systematic investigation of the importance of various parameters on CPET, in addition to models for the key tyrosine-histidine pair in photosystem II.1520 Removal of an electron from these compounds proceeds with concomitant transfer of the phenolic proton to the base. These are not HAT reactions because the electron and proton are separated in the products, so these have been termed separated CPET (or multiple site, bidirectional, or orthogonal EPT).1 In our ongoing studies of phenols with a pendant nitrogen base, the PT transfer distance (dON) has been a parameter of interest, but the kinetic effects of this parameter have been convoluted with changes in other variables. For example, in our study of oxidations of a series of phenol-imidazole compounds, we found that intrinsic barriers to CPET were fairly constant despite variations in ground state dON spanning 0.113(3) Å, as determined by X-ray crystallography.21 However, we concluded that crystal packing forces were likely a major cause of this variation, as computed gas phase values varied by only 0.04 Å with poor correlation to X-ray values. Comparing two phenol-pyridines, we found that the presence of a methylene unit between the phenol and pyridine moieties leads to a longer proton-transfer distance as well as substantially increased intrinsic barriers to CPET.22 This kinetic effect was ascribed to changes in the hydrogen bond, specifically, the disruption of the strong, resonance-assisted hydrogen bond in the fully conjugated case, rather than specifically to the distance.22 This explanation is consistent with kinetic data obtained with phenol-imidazoles which have lower CPET barriers despite longer dON.19b,21 Herein, we describe a detailed experimental study to probe the importance of proton transfer distance on separated CPET, comparing the oxidations of two closely related phenolamine compounds. Reactions of 2,4-di-tert-butyl-6-(diphenylaminomethyl)phenol (1) (Scheme 1) with one-electron, outer-sphere oxidants such as substituted triarylamminium radical cations proceed via CPET to yield the distonic, proton-transferred cation.19a The bicyclic compound 2, 1-amino-4,6-di-tert-butyl-7-hydroxyindan, is an analogue of 1 in which the fused five-membered ring leads

to an increase in dON. The kinetic, thermodynamic, structural, and spectroscopic properties of 1 and 2 are compared below, using both experiment and DFT calculations. The results are not what is predicted by simple theoretical tunneling models, and contrast with those of a paper that appeared just as this one was undergoing final revisions.23 This study can serve as a test case for more sophisticated theories of CPET, and it illustrates the importance of combining experimental and theoretical analyses of PCET.

’ RESULTS Synthesis of Bicyclic Amine 2. The hydrochloride salt of compound 2, 4,6-di-tert-butyl-7-hydroxy-1-aminoindan hydrochloride, has been reported by Oshiro et al.24 The oxime reduction in that report proved problematic in our hands, perhaps a result of the formation of byproducts via ortho-quinone methide intermediates.25 However, 2 was successfully prepared via reductive amination of 7-hydoxy-1-indanone25 followed by alkylation in tert-butanol/H2SO4(aq) in the presence of urea.26 The resulting salt was deprotonated to the neutral amine, 2, which has been characterized by IR, NMR, and mass spectrometry (see Experimental Section). In addition, single crystals of 2 were obtained from a solution in hexanes and were characterized by X-ray diffraction. Structural and Spectroscopic Comparisons of 1 and 2. Compound 2 crystallizes in an orthorhombic space group with one independent molecule per unit cell, with the enantiomers related by symmetry. The triclinic unit cell of 1, however, has two independent molecules;19 the comparisons below either use the average or give the parameters for both molecules explicitly. Figure 1 presents ORTEPs of 2 and one of the independent molecules of 1, as well as the atom labeling used. The structures of 1 and 2 are similar in many ways. The CO distances and comparable bond lengths in the phenol rings of the two compounds are the same within less than 0.02 Å. In both cases, the aliphatic carbon that holds the amino group (Cα) is essentially in the phenol plane, but the amine nitrogen lies out of 17342

dx.doi.org/10.1021/ja2056853 |J. Am. Chem. Soc. 2011, 133, 17341–17352

Journal of the American Chemical Society

ARTICLE

Figure 1. ORTEPs, with crystallographic O 3 3 3 N distances, for phenols 119 and 2; ellipsoids drawn at 75% probability. For 1, the O 3 3 3 N distance listed is the average of the two independent molecules in the unit cell, which have distances that differ by 0.06 Å. Atom labels for relevant carbons used for the geometry discussion are given on the drawings.

the plane of the phenol by 0.827 Å (2) and 0.832, 0.998 Å (1). The twisting of the nitrogen out of the plane is also indicated by the C1C2CαN torsion angles of 35.1(4) (2) and 33.6(3), 41.9(3) (1). In addition, the OHN angles of the hydrogen bonds are the same within uncertainty for 2 (153(3)) and 1 (152(2) and 155(2)). There are also significant differences between the structures. The critical O 3 3 3 N distance (dON) in 2, 2.711(4) Å, is 0.13 ( 0.03 Å longer than in the two independent molecules in the structure of 1, 2.550(2), 2.613(3) Å. The dON in 2 is comparable to that reported for a closely related 7-hydroxy-1-aminoindan derivative, 2.747 ( 0.013 Å (this is the average of two independent molecules in the unit cell).24,27 The longer dON in 2 is due in part to Cα being constrained away by the cyclopentane ring, as indicated by the larger C1C2Cα angle in 2 (113.8(3)) than in 1 (107.3(2) and 107.4(2)). In addition, the NCαC2 angle is 113.8(3) in 2, larger than the tetrahedral angle and larger than the 107.4(3) in 1. The C2Cα and CαN bonds are shorter in 2 than 1, by 0.035(6) Å (C2Cα) and 0.061(5) Å (CαN). These differences may be the result of steric effects in 1 and increased sp2 character at the α-carbon in 2, which is distorted from a tetrahedral geometry; the sum of the three angles, excluding the hydrogen, around Cα is 336.6(5).28 We focus on the O 3 3 3 N distance rather than the OH and H 3 3 3 N distances because the broad vibrational wave function means that the proton does not have a well-defined position. Electron density for the H1 was observed in the structures of 1 and 2 in the difference Fourier map and its position was refined isotropically, but it is not straightforward to interpret this position.29 The crystallographic structures can be significantly influenced by packing forces, as indicated by the 0.063(4) Å difference in dON between the two independent molecules in crystalline 1. Therefore, gas-phase geometries have been computed using Density Functional Theory (DFT), specifically B3LYP/6-31G(d,p). For computational studies, tert-butyl and phenyl groups have been replaced by CH3, with little effect on the value of dON.30 The gasphase DFT structures have longer dON than the crystallographic values, 2.808 Å for 2 and 2.643 Å for 1, but the difference between the two (0.165 Å) is similar to the value obtained from X-ray structures. Additionally, geometry optimizations for 1 and 2 using five expanded basis sets in conjunction with four additional density functionals or the inclusion of a solvent model gave similar results (see Supporting Information). More specifically, in geometries calculated using the largest basis set (aug-cc-pVTZ), dON varies

Figure 2. IR spectra of 1 (blue) and 2 (red) collected as KBr pellets. The lines from the drawings to the spectra indicate the broad νOH band in each spectrum.

by 1/2, the opposite of the observed trend. A change in α might also indicate a change in the nature of the classical transition structure in the reactions of 2.56,57 For example, if the transition structure involved mostly proton transfer with little electron transfer, the barrier would be less sensitive to the redox potential of the oxidant and α should be less than 1/2 when ΔG is varied by changing the outersphere oxidant. This kind of asynchrony has been discussed by Lee, Kreevoy, et al. in the context of hydride transfer reactions,58 and more generally by Bernasconi and others.55,56,59 It is not clear, however, why this would be favored with longer dON or with deuterium substitution. Still, this work and that of Meyer and others indicate that the value of α may provide valuable insight into the nature of the CPET processes and may serve as a benchmark for computational studies.1618,53 Motion along the Proton DonorAcceptor Coordinate. The role of ‘promoting’ or ‘gating’ vibrations in hydrogen transfer reactions—motions which compress the proton transfer distance and thereby facilitate the transfer—has received considerable attention in the literature.1,1215,31,41 The energetics of such motion in compounds 1 and 2 has been examined computationally, using a series of partial geometry optimizations. In each gasphase DFT calculation, dON was fixed and the positions of all of the other atoms were optimized, as described in the Experimental Section. The energies as a function of dON, as shown in Figure 6, are a slice through the multidimentional potential energy surface that describes the CPET reactions. This analysis ignores solvent effects and the presence of the oxidant, but these should be very 17347

dx.doi.org/10.1021/ja2056853 |J. Am. Chem. Soc. 2011, 133, 17341–17352

Journal of the American Chemical Society

ARTICLE

Figure 6. Relative potential energy for partially optimized structures of 1 (blue circles) and 2 (red diamonds) with varied dON. Lines are parabolic fits of the points with dON e dON (optimized).60.

similar for the two phenol-amines reacting with closely related oxidants. Parabolic fits of the points corresponding to compression along dON (Figure 6) give effective force constants of 148 and 92 kcal mol1 Å2 for 1 and 2, respectively.60 Fits to Morse potentials give smaller values in roughly the same ratio (76 and 53 kcal mol1 Å2). We were initially surprised that 2 is easier to compress along the dON coordinate because the bicyclic structure would appear more rigid. However, the dominant factor in determining the magnitude of this force constant appears to be O 3 3 3 N steric repulsion. Still, despite the difference in force constants, it is always more energetically costly for 2 to reach a compressed value of dON than it is for 1. To distort 2 to the calculated dON of ground state 1 requires 1.2 kcal mol1, and to reduce its dON, an additional 0.1 Å requires an additional 1.3 kcal mol1. In a model in which 1 and 2 have to reach the same value of dON to undergo CPET, these energies imply that 2 would react 1 or 2 orders of magnitude slower than 1. This simple calculation applies both to a classical over-the-barrier model and to a proton tunneling model, but only if all of the other parameters important to CPET are equal for 1 and 2. In sum, the similarity of their CPET reactivity of 1 and 2 does not follow in a straightforward manner from ground state dON and changes in the donoracceptor motions. A Deeper Look: Proton Vibrational Wave Functions. In the multistate continuum theory for PCET developed by HammesSchiffer and co-workers, one set of significant kinetic parameters is the overlaps of reactant and product vibrational wave functions (FranckCondon factors). As part of a more extensive study,48 we have calculated these wave functions (with certain assumptions) for 1 and 2, as shown in Figure 7. Multistate continuum theory is conceptually related to Marcus theory, in which the nontransferring atoms and solvent reorganize to a surface crossing point where the electron/proton tunneling occurs. Our calculations fixed all of the atoms at a structure halfway between the phenol and the product radical cation, termed the TS geometry, then determined the energy as a function of proton position between the O and the N. The derived surface was fitted to a sixth-order polynomial, which was used to derive the proton wave functions (see Experimental Section). The neutral phenols at the TS geometry have a single-well, distorted harmonic potential for the proton bound to the phenol oxygen (Figure 7, bottom). The zwitterionic form with the proton bound to the nitrogen, OArNH3+, is high in energy, >15 kcal mol1 for both compounds 1 (left) and 2 (right). In contrast, the distonic radical cations 1•+ and 2•+ have the proton potential energy surfaces at the TS geometry that are double minima (top plots). The ground vibrational states (blue lines) are localized on the nitrogen, on the right, but the first excited

Figure 7. Proton transfer potential energy surfaces (black lines) calculated for neutrals (bottom) and distonic radical cations (top) of 1 and 2, with heavy atoms fixed at the average “TS0” geometry, dON = 2.59 Å (1) and 2.67 Å (2). Proton vibrational wave functions are shown: ν = 0 (blue) and ν = 1 (red); higher vibrational wave functions (dotted lines) are minor contributors to the CPET reaction.

states (red) have more OH bond character. The bicyclic compound 2•+ has an especially pronounced double minimum due to its longer ON distance, and the ν = 1 state is almost fully localized at oxygen. The FranckCondon overlap between the ground ν = 0 state in 2 and the ν = 1 state in 2•+ is almost unity, S0,12 = Æϕ0(2)|ϕ1(2•+)æ2 = 0.90. In addition, this first vibrational excited state is calculated to be only 1.3 kcal mol1 or 450 cm1 above the ground state, so the energetic cost to access this state is small. For compound 1 at its TS geometry, the first excited state is two times higher (935 cm1) and the overlap is smaller, S0,12 = 0.54. This vibrational analysis suggests a possible explanation for the similarity of the CPET reactions of compounds 1 and 2. The higher classical barrier for CPET in 2 due to the longer distance could be balanced by the facility of reaction from the ν = 0 state in 2 to the ν = 1 state in 2•+, ϕ0(2) f ϕ1(2•+). This is more facile for 2 because the longer distance makes the ϕ1(2•+) lower in energy and more localized on the oxygen. However, structures with smaller dON probably contribute significantly to the rate constants, and a complete study including the vibrational overlaps for structures with a range of O 3 3 3 N distances will be required to indicate if these trends are indeed key factors in the kinetics of 1 and 2. Vibrational excited states should be even more important in deuterium transfers, so this analysis also provides a qualitative rationalization for the counterintuitive lower KIEs for compound 2. While the magnitude of the KIE differences vary with driving force, the trend remains unchanged, KIE(1) > KIE(2). This is opposite to the prediction of a simple tunneling model, in which the XD vibrational wave functions penetrate less into the forbidden region so that at longer distances tunneling is less favored and KIEs are larger. As Hammes-Schiffer et al. state in their analysis of KIEs for CPET reactions, “typically the KIE will increase with increasing equilibrium proton donoracceptor distance if all other parameters remain fixed.”40 That study 17348

dx.doi.org/10.1021/ja2056853 |J. Am. Chem. Soc. 2011, 133, 17341–17352

Journal of the American Chemical Society shows, however, that there are a number of situations where an inverse relationship of KIE and PT distance can occur, for instance when the species with a longer distance has a lower frequency for the vibration along the proton donoracceptor coordinate.40 Experimental support for this case has been provided by high pressure stopped flow and site directed mutagenesis studies on hydride transfer reactivity of morphinone reductase by Scrutton and co-workers.14c,d For the phenols presented here, the effective frequency of this motion is lower in 2 than in 1 (Figure 6), so this effect could also be a cause of the decreased KIEs observed in reactions of 2. The involvement of vibrational excited states and the differences in effective dON vibrational frequencies are likely to also be the origins of the unusual effects of driving force on the reactions of 1 and 2. For instance, the KIEs increase with ΔGCPET, in contrast with the semiclassical prediction that the KIE should be a maximum at ΔGreaction = 0. Reactions of 2 show an unusually shallow dependence of rate constant on driving force, with the Brønsted α [Δln(k)/Δln(Keq)] = 0.38(3). The reactions of deuterated 2 show an even lower α of 0.33. These values are significantly below the typical α = 1/2 found for 1 and related phenol-base compounds.19,21,22,35b As discussed above, the deviation of α from 1/2 does not come from the parabolic dependence of rate on driving force from the Marcus equation (eq 4). It must instead come from a more subtle dependence of other vibrational and quantum coupling factors, unique to CPET, on ΔGCPET.

’ CONCLUSIONS This study compares CPET oxidations of two closely related phenol-amines, phenol-amine 1 and bicyclic amino-indanol 2, in which the proton donoracceptor distance varies by roughly 0.15 Å. The difference in distances is shown to be more accurately determined by computational studies rather than crystallographic results, because of the effects of crystal packing on the lowfrequency donoracceptor mode. These results provide valuable new test cases for the development of theory, which, in turn, will guide new experiments and the interpretation of kinetic and structural chemical and biochemical data. In contrast with the results of both classical and simple tunneling models, the kinetic effects of the different donoracceptor distances in 1 and 2 are small. For reactions with similar oxidants when ΔGCPET is near zero, the rate constants for 1 and 2 are very similar, within a factor of ca. 2. The H/D kinetic isotope effects for the reactions of 2, with the longer proton-transfer distance, are smaller than those observed for 1, and increase with increasing ΔGCPET. Reactions of 2 and 2-d3 show an unusually shallow dependence of rate on driving force, with Brønsted α [Δln(k)/ Δln(Keq)] values of 0.38(3) and 0.33, respectively. These results contrast with the conclusions of a very recent report, that phenolbase compounds with (crystallographically) shorter, conjugated hydrogen bonds react faster than those with nonconjugated H-bonds.23 Future studies will determine whether that conclusion is biased by comparing compounds with conjugated versus nonconjugated hydrogen bonds (as suggested by earlier studies22,31) or whether the phenol-amine compounds 1 and 2 compared here are atypical for some reason. Classical, over-the-barrier models predict much slower rates for 2 due to the larger barrier to proton transfer at larger distances and to the larger energetic costs to reach shorter proton donor acceptor distances. First-order tunneling models also predict

ARTICLE

slower rates and higher kinetic isotope effects for 2 because of the longer tunneling distance. Simple tunneling models assume (1) tunneling from ν = 0 in the reactant to ν = 0 in the product, (2) small overlap between the vibrational wave functions, and (3) Morse-like potentials for the protons. The vibrational analysis summarized in Figure 7 shows that none of these three assumptions hold for 1 and 2. Simple tunneling models are not appropriate for strongly hydrogen-bonded systems such as these because the proton potential surfaces are highly anharmonic and often exhibit double minima. The surfaces also change substantially at shorter donoracceptor distances,61 structures which often contribute significantly to the CPET rate constant.62 The experimental and computational comparison of 1 and 2 indicates that changing the proton-donor acceptor distance in hydrogen bonded systems affects multiple parameters that are important to CPET. Changing the distance strongly affects the shape of the proton potential energy surface and therefore the vibrational overlaps involving vibronic excited states. Compounds with a longer donoracceptor distance are also likely to have lower effective force constants for distortion along this coordinate. These effects are not evident from simple tunneling models and illustrate the complexity of CPET processes.

’ EXPERIMENTAL SECTION General Considerations. Unless otherwise noted, reagents were purchased from Aldrich, solvents from Fisher, and deuterated solvents from Cambridge Isotope. MeCN was used as obtained from Burdick and Jackson (low-water brand) and stored in an argon-pressurized stainless steel drum plumbed directly into a glovebox. CD3CN was dried by stirring over CaH2 overnight, then distilled onto P2O5, followed by another distillation from CaH2 and stored in a glovebox. Ammonium acetate was dissolved in methanol and evaporated to dryness (3) and the residue was dried in vacuo at 40 C for 24 h. n Bu4NPF6 was recrystallized three times from EtOH and dried in vacuo for 2 days at 110 C prior to use. The iron-bipyridyl and phenanthroline complexes were synthesized according to literature procedures63 and were used as PF6 salts. Triarylamminium salts were prepared from the corresponding amines as described previously.19b 1 H NMR and 13C NMR spectra were recorded on Bruker AV300, AV301, DRX499, or AV500 spectrometers at ambient temperatures; chemical shifts are reported relative to TMS in ppm by referencing to the residual solvent signals. The UVvis spectra were obtained on a Hewlett-Packard 8453 diode array spectrophotometer. Infrared spectra (Figure 2) were obtained from KBr pellets on a Bruker Vector 33. Mass spectrometry was performed on a Kratos Profile HV-3 direct inlet, electron impact mass spectrometer. Calculations. Unless stated otherwise, all calculations were performed using Gaussian 03.64 In these computational studies, tert-butyl groups were replaced by CH3. Geometry optimizations, IR frequency calculations, and potential energy scans of the ON coordinate utilized B3LYP/6-31G(d,p) level of theory. Calculation of the proton transfer surfaces utilized the (U)B3LYP functional with 6-31G(d) basis set used for all atoms with the exception of the transferring proton, for which a set of p polarization functions were included, that is, 6-31G(d,p). All calculations were performed in the gas phase. PCM models of MeCN solvent with ε = εop appear to have relatively minor effects on the shape of the hydrogen bond potentials for the neutral and cation. Proton vibrational energy levels and wave functions were calculated numerically for these one-dimensional potential energy wells using a basis set of 200 harmonic oscillator functions.65 Synthesis. 1-Amino-4,6-di-tert-butyl-7-hydroxyindan (2). tertButanol (4 mL) was added to a stirred solution of 1-amino-7-hydroxyindan hydrochloride24,25 (0.409 g, 2.28 mmol) and urea (0.398 g) in 17349

dx.doi.org/10.1021/ja2056853 |J. Am. Chem. Soc. 2011, 133, 17341–17352

Journal of the American Chemical Society 75% H2SO4 (aq). Beige solids formed as the reaction mixture was stirred at ambient temperatures for 52 h. Ice (15 g) was added and the suspension was filtered while cold. The solids were washed with cold water (15 mL) and dried thoroughly in vacuo. The residue was suspended in 15% aqueous methanol and the pH was adjusted to 9 with the addition of NaOH(aq). This solution was extracted with methylene chloride (3  30 mL) and the combined organic layers were washed with water and brine, then dried over MgSO4. Removal of solvent on the rotary evaporator gave a beige solid (0.55 g). Recrystallization from boiling hexanes gave 2 as a pale yellow crystalline solid (0.348 g, 61%). 1H NMR (CD3CN) (using IUPAC indan numbering): 1.30 (9 H, s, C(CH3)3), 1.37 (9 H, s, C(CH3)3), 1.60 (1 H, m, 2-H), 2.10 (2 H, br, NH2), 2.46 (1 H, m, 2-H), 2.74 (1 H, m, 3-H), 3.10 (1 H, m, 3-H), 4.36 (1 H, m, 1-H), 7.08 (1 H, s, 5-H), 11.14 (1 H, br, OH). MS (m/z): 261 (M+). The rate constant for OH/NH2 exchange was determined from the EXSY spectra using the relation: k = tm1 ln(r + 1/r  1), where tm is the mixing time and the term r is related to the intensity of the cross and diagonal peaks (IAB, IAA, and IBB), r = 4XAXB(IAA + IBB)/2IAB  (XA  XB)2, where X is the mole fraction population of a given site.66 No correction for an NOE was made. Cyclic Voltammetry. Cyclic voltammograms were taken on an E2 Epsilon electrochemical analyzer (Bioanalytical Systems) at ca. 5 mM substrate in anaerobic 0.1 M nBu4NPF6 acetonitrile solution. The electrodes were as follows: working, Pt disc; auxiliary, platinum wire; and reference, Ag/AgNO3 (0.01 M) in electrolyte solution. The working electrode was polished with alumina, then rinsed with water, dilute HNO3(aq), water, and ethanol following each scan. All potentials are reported versus a Cp2Fe+/0 internal standard. Kinetics. Kinetics experiments were performed using an OLIS RSM1000 stopped-flow in anaerobic MeCN. The data were analyzed with SpecFit global analysis software.67 For the reaction of 2 + 3e+, rates were also obtained by fitting the absorbance at a single wavelength using Microsoft Excel as was done for the reaction of 1 + 3e+.19 The rates obtained via these two methods were the same within error. Kinetics were fit to pseudo-first-order, second-order, or opposing second-order kinetics as appropriate. Kinetics for the reactions utilizing iron(III) oxidants were obtained at 0.1 M nBu4NPF6 as the potentials for these compounds vary with ionic strength due to ion pairing.68 Typical conditions used for reactions 1 and 2 were [3]i ≈ 1.5  105 M and 440 equiv of 2 or 330 equiv of 1. The variation of the optical extinction coefficients (ε with temperature of 3a+ and 3b+ was determined. Two solutions each of 3a+ and 3b+ were taken up in temperature from ca. 299 to 327 K and the absorbance at the λmax (718 nm for 3a+ and 748 nm for 3b+) was recorded at six temperatures. At each temperature, ε at λmax of the amminium was taken as the slope of the best fit line in plots of absorbance versus [3], with the inclusion of zero as a point. For each amminium, the variation of ε (M1 cm1) versus T (K) is linear. For 3a+, ε = 102T + 31 300 (R2 = 0.998); for 3b+, ε = 103T + 59 400 (R2 = 0.994). Crystal Structure of 2. A colorless block 0.12  0.10  0.04 mm in size was mounted on a Cryoloop with Paratone oil. Data were collected in a nitrogen gas stream at 208(2) K using phi and omega scans. Crystalto-detector distance was 60 mm and exposure time was 30 s per frame using a scan width of 0.5. Data collection was 99.7% complete to 25.00 in θ. A total of 19 339 reflections were collected covering the indices, 11 e h e 10, 11 e k e 14, 32 e l e 31. In total, 2794 reflections were found to be symmetry independent, with an Rint of 0.1033. Indexing and unit cell refinement indicated a primitive, orthorhombic lattice. The space group was found to be Pbca (No. 61). The data were integrated using the Bruker SAINT software program and scaled using the SADABS software program. Solution by direct methods (SIR-2004) produced a complete heavy-atom phasing model consistent with the proposed structure. All non-hydrogen atoms were refined anisotropically by full-matrix least-squares (SHELXL-97). All hydrogen atoms, with the exception of the hydroxyl hydrogen H1, were placed using a

ARTICLE

riding model. Their positions were constrained relative to their parent atom using the appropriate HFIX command in SHELXL-97. The hydroxyl hydrogen H1 was located from the difference map and its position was refined isotropically.

’ ASSOCIATED CONTENT

bS

Supporting Information. Complete citation for ref 64, 2-D EXSY spectrum, solution IR spectra of 1 and 1-d3, details on the spectrophotometric determination of equilibrium constants and the crystal structure of 2. This material is available free of charge via the Internet at http://pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

[email protected]; [email protected] Present Addresses

§ Uppsala University, Department of Photochemistry and Molecular Science, Box 523, S-75120 Uppsala, Sweden. † Albright College, Department of Chemistry and Biochemistry, P.O. Box 15234, Reading, PA 19612-5234; (e-mail) [email protected].

’ ACKNOWLEDGMENT We thank Dr. Antonio DiPasquale for the X-ray crystal determination, Drs. Abraham Nudelman and Yaacov Herzig for providing details on the preparation of 7-hydroxy-1-aminoindan, Drs. Eric Heatwole, Andri Arnaldsson, and Oleg Prezhdo for sharing their program to calculate vibrational wave functions from sixthorder polynomial surfaces, and Dr. Adam Tenderholt for comments on the manuscript and some additional computations. We are grateful to the U.S. National Institutes of Health (GM50422), and the University of Washington for support of this work. ’ REFERENCES (1) (a) Huynh, M. H. V.; Meyer, T. J. Chem. Rev. 2007, 107, 5004–5064. (b) Hammes-Schiffer, S. Chem. Rev. 2010, 110, 69376938 (Introduction to a special issue of Chem. Rev. on PCET). (c) Hodgkiss, J. M.; Rosenthal, J.; Nocera, D. G. In Hydrogen-Transfer Reactions; Hynes, J. T.; Klinman, J. P., Limbach, H.-H., Schowen, R. L., Eds.; Wiley-VCH: Weinheim, 2006; pp 503562. (d) Warren, J. J.; Tronic, T. A.; Mayer, J. M. Chem. Rev. 2010, 110, 6961–7001. (e) Costentin, C. Chem. Rev. 2008, 108, 2145–2179. (2) Pu, J.; Gao, J.; Truhlar, D. G. Chem. Rev. 2006, 106, 3140–3169. (3) Klinman, J. P. Biochim. Biophys. Acta 2006, 1757, 981–987. (4) (a) Stubbe, J.; van der Donk, W. A. Chem. Rev. 1998, 98, 705–762. (b) Pesavento, R. P.; van der Donk, W. A. Adv. Protein Chem. 2001, 58, 317. (5) The term ‘CPET’ was coined in Costentin, C.; Evans, D. H.; Robert, M.; Saveant, J.-M.; Singh, P. S. J. Am. Chem. Soc. 2005, 127, 1249012491. (6) (a) Marcus, R. A.; Sutin, N. Biochim. Biophys. Acta 1985, 811, 265–322. (b) Barbara, P. F.; Meyer, J. T.; Ratner, M. A. J. Phys. Chem. 1996, 100, 13148–13168. (7) cf. (a) Mayer, J. M. Acc. Chem. Res. 2011, 44, 36–46. (b) Bonin, J.; Costentin, C.; Louault, C.; Robert, M.; Saveant, J.-M. J. Am. Chem. Soc. 2011, 133, 66686674 and references therein. (c) Mayer, J. M. J. Phys. Chem. Lett. 2011, 2, 1481–1489. (8) (a) Hammes-Schiffer, S. In Hydrogen-Transfer Reactions; Hynes, J. T., Klinman, J. P., Limbach, H. H., Schowen, R. L., Eds.; Wiley-VCH: Weinheim, 2007; pp 479503. (b) Hammes-Schiffer, S.; Alexei A. Stuchebrukhov, A. A. Chem. Rev. 2010, 110, 6939–6960. (c) Marcus, R. A. J. Phys. Chem. B 2007, 111, 6643–6654. (9) Krishtalik, L. I. Biochim. Biophys. Acta 2000, 1458, 6–27. 17350

dx.doi.org/10.1021/ja2056853 |J. Am. Chem. Soc. 2011, 133, 17341–17352

Journal of the American Chemical Society (10) (a) Rickert, K. W.; Klinman, J. P. Biochemistry 1999, 38, 12218–12228. (b) Liang, Z.-X.; Klinman, J. P. Curr. Opin. Struct. Bio. 2004, 14, 648–655. (c) Lewis, E. R.; Johansen, E.; Holman, T. R. J. Am. Chem. Soc. 1999, 121, 1395–1396. (d) Costas, M.; Mehn, M. P.; Jensen, M. P.; Que, L. Chem. Rev. 2004, 104, 939–986. (11) cf. (a) Hatcher, E.; Soudackov, A. V.; Hammes-Schiffer, S. J. Am. Chem. Soc. 2004, 126, 5763–5775. (b) Lehnert, N.; Solomon, E. I. J. Biol. Inorg. Chem. 2003, 8, 294–305. (c) Meyer, M. P.; Klinman, J. P. J. Am. Chem. Soc. 2011, 133, 430–439. (12) (a) Knapp, M. J.; Klinman, J. P. J. Am. Chem. Soc. 2002, 124, 3865–3874. (b) Klinman, J. P. Philos. Trans. R. Soc., B 2006, 361, 1323–1331. (c) Meyer, M. P.; Klinman, J. P. J. Am. Chem. Soc. 2011, 133, 430–439. (d) Meyer, M. P.; Klinman, J. P. Chem. Phys. 2005, 319, 283–296. (13) (a) Hatcher, E.; Soudackov, A. V.; Hammes-Schiffer, S. J. Am. Chem. Soc. 2004, 126, 5763–5775. (b) Hatcher, E.; Soudackov, A. V.; Hammes-Schiffer, S. J. Am. Chem. Soc. 2007, 129, 187–196. (14) cf. (a) Masgrau, L.; Roujeinikova, A.; Johannissen, L. O.; Hothi, P.; Basran, J.; Ranaghan, K. E.; Mulholland, A. .; Sutcliffe, M. J.; Scrutton, N. S.; Leys, D. Science 2006, 312, 237. (b) Johannissen, L. O.; Hay, S.; Scrutton, N. S.; Sutcliffe, M. J. J. Phys. Chem. B 2007, 111, 2631–2638. (c) Hay, S.; Sutcliffe, M. J.; Scrutton, N. S. Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 507. (d) Pudney, C. R.; Johannissen, L. O.; Sutcliffe, M. J.; Hay, S.; Scrutton, N. S. J. Am. Chem. Soc. 2010, 132, 11329–11335. (e) Peters, B. J. Chem. Theory Comput. 2010, 6, 1447–1454. (f) Benkovic, S. J.; Hammes-Schiffer, S. Science 2003, 301, 1196–1202. (15) (a) Biczok, L.; Gupta, N.; Linschitz, H. J. Am. Chem. Soc. 1997, 119, 12601–12609. (b) Gupta, N.; Linschitz, H. J. Am. Chem. Soc. 1997, 119, 6384–6391. (16) (a) Sj€odin, M.; Styring, S.; Wolpher, H.; Xu, Y.; Sun, L.; Hammarstr€om, L. J. Am. Chem. Soc. 2005, 127, 3855–3863. (b) Sj€odin, M.; Irebo, T.; Utlas, T. E.; Lind, J.; Merenyi, G.; Åkermark, B.; Hammarstr€om, L. J. Am. Chem. Soc. 2006, 128, 13076–13083. (c) Irebo, T.; Reece, S. Y.; Sj€odin, M.; Nocera, D. G.; Hammarstr€om, L. J. Am. Chem. Soc. 2007, 129, 15462–15464. (d) Irebo, T.; Johansson, O.; Hammarstr€om, L. J. Am. Chem. Soc. 2008, 130, 9194–9195. (e) Sj€odin, M.; Styring, S.; Åkermark, B.; Sun, L.; Hammarstr€om, L. J. Am. Chem. Soc. 2000, 122, 3932–3936. (17) Reece, S. Y.; Nocera, D. F. J. Am. Chem. Soc. 2005, 127, 9448– 9458. (18) (a) Costentin, C.; Robert, M.; Saveant, J.-M. J. Am. Chem. Soc. 2006, 128, 8726–8727. (b) Costentin, C.; Robert, M.; Saveant, J.-M. J. Am. Chem. Soc. 2007, 129, 9953–9963. (c) Correction to reference 18b: Costentin, C.; Robert, M.; Saveant, J.-M. J. Am. Chem. Soc. 2010, 132, 2845. (d) Bonin, J.; Costentin, C.; Louault, C.; Robert, M.; Routier, M.; Saveant, J.-M. Proc. Natl. Acad. Sci. U.S.A. 2010, 107, 3367–3372. (19) (a) Rhile, I. J.; Mayer, J. M. J. Am. Chem. Soc. 2004, 126, 12718–12719. (b) Rhile, I. J.; Markle, T. F.; Nagao, H.; DiPasquale, A. G.; Lam, O. P.; Lockwood, M. A.; Rotter, K.; Mayer, J. M. J. Am. Chem. Soc. 2006, 128, 6075–6088. (20) Song, N.; Stanbury, D. M. Inorg. Chem. 2008, 47, 11458–11460. (21) Markle, T. F.; Rhile, I. J.; DiPasquale, A. G.; Mayer, J. M. Proc. Natl. Acad. Sci. U.S.A. 2008, 105, 8185–8190. (22) Markle, T. F.; Mayer, J. M. Angew. Chem., Int. Ed. 2008, 47, 738–740. (23) Zhang, M.-T.; Irebo, T.; Johansson, O.; Hammarstr€om, L. J. Am. Chem. Soc. 2011, 133, 13224–13227. (24) Oshiro, Y.; Sakurai, Y.; Tanaka, T.; Ueda, H.; Kikuchi, T.; Tottori, K. J. Med. Chem. 1991, 34, 2004–2013. (25) Herzig, Y.; Lerman, L.; Goldenberg, W.; Lerner, D.; Gottlieb, H. E.; Nudelman, A. J. Org. Chem. 2006, 71, 4130–4140. (26) Nevrekar, N. B.; Sawardekar, S. R.; Pandit, T. S.; Kudav, N. A. Chem. Ind. 1983, 206–207. (27) The structure for 1-amino-6-ehtyl-7-hydroxy-4-methyl-2-phenylindan reported in reference 24 is available at Cambridge Structural Database, reference code KOFNOI. (28) The calculated Mayer Bond orders for the bonds to Cα in 1 (the calculated Cα-gem-diemthyl analogue) and 2 are consistent with a

ARTICLE

small difference in hybridization: CαN, 0.913 (1) vs 0.940 (2); CαNC2, 0.946 (1) vs 0.956 (2); CαCβ, 1.003 (1) vs 0.974 (2); CαC/H, 0.991 (1, CαCH3) vs 0.935 (2, CαH). (29) Some of the complexity in structural studies of strong hydrogen bonds are evident in: Fontaine-Vive, F.; Johnson, M. R.; Kearley, G. J.; Howard, J. A. K.; Parker, S. F. J. Am. Chem. Soc. 2006, 128, 29632969. (30) The optimized dON value for 1 with two phenyl groups on Cα (2.633 Å) is essentially identical to that of the simplified Cα-gemdimethyl analogue (dON = 2.640 Å). (31) Johannissen, L. O.; Irebo, T.; Sj€odin, M.; Johansson, O.; Hammarstr€ om, L. J. Phys. Chem. B. 2009, 113, 16214–16225. (32) In ref 31, Hammarstr€om, et al. noted that in two phenolcarboxylates which differ by a methylene unit, the donoracceptor distance varied by 0.05 Å; however, the difference in the proton tunneling distance was identical within 0.01 Å because the longer hydrogen bond was also more linear. This is likely true in other studies where the dON is varied by insertion of a methylene unit between phenol and base (i.e., refs 22 and 23). (33) Bratos, S.; Leicknam, J.-C.; Gallot, G.; Ratajczak, H. In Ultafast Hydrogen Bonding Dynamics and Proton Transfer Processes in the Condensed Phase; Elsaaesser, T., Bakker, H. J., Eds.; Kluwer Academic: Boston, MA, 2002; pp 530. (34) Jeffrey, G. A. An Introduction to Hydrogen Bonding: Oxford University Press: New York, 1997, pp 220225. (35) Harmonic vibrational analyses performed at B3LYP/6-31G(d,p) and corrected following Korth et al.35a predict νOH to be 193 cm1 lower for 1 compared to 2 (3068 and 3261 cm1, respectively). However, the OHN surfaces are highly anharmonic and the analysis of hydrogen bond vibrations is not straightforward, so this calculation would not be expected to give quantitative agreement. (a) Korth, H.-G.; de Heer, M. I.; Mulder, P. J Phys Chem A 2002, 106, 8779–8789. ν(OH)DFT,corr = [ν(OH)DFT  159.5]  0.9904. (b) Markle, T. F. Concerted ProtonElectron Transfer in the Oxidation of Hydrogen Bonded Phenol-Base Compounds, Ph.D. Thesis, University of Washington, 2009. (c) ref 48a. (36) Rhile, I. J.; Mayer, J. M. Angew. Chem., Int. Ed. 2005, 44, 1598–1599. (37) In ref 19b, rates were determined for the reaction 1 + 3c+ with T = 279.6327.0 K and fit to the Eyring equation yielding ΔHq = 6.3 ( 0.8 kcal mol1 and ΔSq = 14.1 ( 2 eu. Fitting this same data to the Arrhenius equation gives Ea = 6.9 ( 0.8 kcal mol1 and log A = 10.2 ( 0.6. (38) Bell, R. P. The Proton in Chemistry, 2nd Ed.; Cornell University Press: Ithaca, NY, 1973; pp 226249. (39) Bell, R. P. The Tunnel Effect in Chemistry; Chapman and Hall: New York, 1980; pp 77105. (40) Edwards, S. E.; Soudackov, A. V.; Hammes-Schiffer, S. J. Phys. Chem. A. 2009, 113, 2117–2126. (41) (a) Keifer, P. M.; Hynes, J. T. J. Phys. Chem. A. 2004, 108, 1179–11808. (b) Keifer, P. M.; Hynes, J. T. J. Phys. Chem. A. 2004, 108, 11809–11818. (42) (a) Excluding 2-d3 + 3e+, the most downhill reaction, from the linear fit gives an excellent correlation, αD = 0.30(3), R2 = 0.997. Extrapolating this line predicts a KIE of ∼10 for the 2 + 3e+ reaction. (b) Because the 2 + 3e+ reaction is highly exoergic (Erxn = 0.43 V), the stepwise ET/PT pathway may compete with CPET, particularly for the reaction of 2-d3 since its KIE will disfavor CPET relative to simple ET. Estimates of the energetics suggest that outer-sphere ET from 2 to 3e+ to yield the phenol radical cation (+•HOArNH2) is a possible if somewhat unlikely initial step.42c,35b If this pathway is contributing in the deuterium case, it would lead to a higher than expected kD and decreased KIE. (c) The ΔGET involves the redox potential of for outer-sphere electron removal from 2 without proton motion, which is not experimentally accessible. This potential is estimated to be close to that of tri-tertbutylphenol, ∼1.09 V,19b which makes ΔGET = 0.42 V. (43) The reaction barrier cannot be smaller than the ΔG. The reactions may not be well described by the Eyring equation, for instance, because nonadiabatic effects will reduce the prefactor below kBT/h, but ΔGq is a lower limit to the actual barrier height. (44) As noted, α for the reactions of 2 + 3+ is significantly lower than 1/2; this is further evidence against the ET/PT mechanism. 17351

dx.doi.org/10.1021/ja2056853 |J. Am. Chem. Soc. 2011, 133, 17341–17352

Journal of the American Chemical Society

ARTICLE

(45) Kuznetsov, A. M.; Ulstrup, J. Can. J. Chem. 1999, 77, 1085–1096. (46) Nagel, Z. D.; Klinman, J. P. Nat. Chem. Biol. 2009, 5, 543–550. (47) For instance, one of the compounds discussed in ref 23, a conjugated phenol-benzimidazole, has four independent molecules in two crystal structures, with dON values that vary by 0.075 Å (2.533(3)2.6087(19)).21,47a Also in ref 23, the structure of the phenol-CH2-benzimidazole appears to have an unusual geometry for the key OH 3 3 3 N hydrogen bond due to an additional intermolecular N 3 3 3 HN hydrogen bonding interaction; this points the basic nitrogen somewhat away from the phenolic OH leading to a longer apparent dON. (a) Benisvy, L.; Bill, E.; Blake, A. J.; Collison, D.; Davies, E. S.; Garner, C. D.; McArdle, G.; McInnes, E. J. L.; McMaster, J.; Ross, S. H. K.; Wilson, C. Dalton Trans. 2006, 258–267. (48) (a) Markle. T. F.; Tenderholt, A. L.; Mayer, J. M. submitted. (b) Ref 35b, Chapter 4. (c) As described in these reports, the proton potential energy surfaces are highly anharmonic and vary substantially between different classes of molecules and with proton donoracceptor distance, suggesting that the vibrational coupling terms will not simply follow a exponential dependence. (49) Ref 6. ΔG* is the free energy barrier, λ is the intrinsic barrier, k is the transmission coefficient, and ν is a generalized frequency factor (typically ∼1013 s1, kBT/h in the Eyring equation). (50) Recasting the data found in ref 22 as ln k versus ln Keq gives a linear plot with a slope = 0.53 and R2 = 0.992. (51) When ln k correlates linearly with ΔG for a set of reactions, as observed here, it is not possible to directly determine both λ and kυ. The Saveant derivation of λ requires a number of simplifications and external calculations. The linearization of the Marcus equation which requires α = 1/2 is clear in eq 24 of ref 18b. (52) When pathways involving vibrational excited states are important, ΔG in eq 3 becomes ΔG + εrel,μν, where εrel,μν is the change in ΔG for a pathway from vibronic states μ to ν.8 It then follows that ∂ΔG*/∂ΔG = 1/2 + (ΔG + εrel,μν)/2λ. (53) Fecenko, C. J.; Thorp, H. H.; Meyer, T. J. J. Am. Chem. Soc. 2007, 129, 15098–10599. (54) Madhiri, N.; Finklea, H. O. Langmuir 2006, 22, 10643–10651. (55) Ludlow, M. K.; Soudackov, A. V.; Hammes-Schiffer, S. J. Am. Chem. Soc. 2010, 132, 1234–1235. (56) Jencks, W. P. Chem. Rev. 1985, 85, 511527, and references therein. (57) Williams, I. H. In Hydrogen-Transfer Reactions; Hynes, J. T., Klinman, J. P., Limbach, H.-H., Schowen, R. L., Eds.; Wiley-VCH: Weinheim, 2006; pp 583602. (58) Lee, I.-S. H.; Chow, K.-H.; Kreevoy, M. M. J. Am. Chem. Soc. 2002, 124, 7755–7761. (59) Bernasconi, C. F. J. Phys. Org. Chem. 2004, 17, 951–956. (60) The parabolic and Morse potential fits only used the energies from structures with dON e dON (optimized) because structures with longer dON distorted torsionally in a way that is not related to the energy to simply distort along dON, and that is not related to the CPET process. (61) See for example:(a) Doslic, N.; K€uhn, O. Zeit. Phys. Chem. 2003, 217, 1507–1524. (b) Jezierska, A.; Panek, J. J.; Koll, A.; Mavri, J. J. Chem. Phys. 2007, 126, 205101. (62) Hammes-Schiffer, S. Acc. Chem. Res. 2006, 39, 93–100. (63) DeSimone, R. E.; Drago, R. S. J. Am. Chem. Soc. 1970, 92, 2343–2352. (64) Frisch, M. J.; et al. Gaussian 03, Revision D.02, Gaussian, Inc.: Wallingford, CT, 2004. For complete citation see Supporting Information. (65) Using code generously provided by Heatwole, E.; Arnaldsson, A.; Prezhdo, O., unpublished work. (66) Perrin, C. L.; Dwyer, T. J. Chem. Rev. 1990, 90, 935–967. (67) Binstead, R. A.; Zuberb€uhler, A. D.; Jung, B. Specfit, version 3.0.36 (32-bit Windows); Spectrum Software Associates: Chapel Hill, NC, 2004. (68) (a) Noel, M.; Vasu, K. I. Cyclic Voltammetry and the Frontiers of Electrochemistry; Aspect:: London, 1990; pp 141143. (b) Braga, T. G.; Wahl, A. C. J. Phys. Chem. 1985, 89, 5822–5828. (c) Chan, M.-S.; Wahl, A. C. J. Phys. Chem. 1978, 82, 2542–2549.

17352

dx.doi.org/10.1021/ja2056853 |J. Am. Chem. Soc. 2011, 133, 17341–17352